Safety of Proton Pump Inhibitors: Current Evidencefor Osteoporosis and Interaction with Antiplatelet Agents

David A. Johnson

Published online: 24 April 2010# Springer Science+Business Media, LLC 2010

Abstract Proton pump inhibitors (PPIs) are commonlyused in millions of patients. Despite the widespread usageand favorable safety profile, much attention has focusedrecently on potential adverse events associated with use.One area of particular concern is the reported potentialadverse effects on bone density, causing possible osteopo-rosis and related bone fractures. The low magnitude of theassociation (OR<2), lack of experimental evidence doc-umenting a mechanism, and inability to assess potentialconfounding factors limit statements regarding causality.The other area of particular concern has been the combineduse with antiplatelet agents, in particular with clopidogrel.Several retrospective studies suggested an adverse cardio-vascular outcome if clopidogrel was combined with PPIs.The theoretic concern was raised to a possible interferencewith the bioactivation of the prodrug via competitiveinterference for both drugs at the cytochrome P450pathway. Although ex vitro studies suggested a possibleeffect and several retrospective analyses were supportive ofan adverse clinical outcome, the more recent prospectivecontrolled trial data do not support an adverse cardiovas-cular outcome. This report reviews the scientific hypothe-sis, biologic plausibility, and the most recent clinical trialdata for both areas of controversy.

Keywords Proton pump inhibitor . Antiplatelet agents .

Osteoporosis . Clopidogrel

Introduction

In the past few years, data have accumulated on thepotential risks of long-term use of proton pump inhibitors(PPIs) in patients with gastroesophageal reflux disease(GERD). The reports were initially sporadic, but a streamof publications has claimed a variety of adverse effects thatmay afflict patients who are chronically treated with PPIs,raising concerns among physicians and patients. Inresponse to such publications, many physicians have alteredtheir usual approach to the care of patients on chronicantireflux treatment for GERD. This review focuses on theliterature related to the recent controversies with PPIs andissues of osteoporosis/hip fracture as well as interactionswith antiplatelet therapies (in particular, clopidogrel).

PPI and Related Osteoporosis/Hip Fracture

Scientific Hypothesis for Causality

Recent epidemiologic studies suggested an associationbetween PPI use and hip fracture [1–3]. The mechanismfor this result most commonly was suggested to be relatedto diminished acid leading to inhibition of the intestinalabsorption of calcium, and hence an accelerated bonemineral loss and subsequent osteoporosis and relatedfracture complications.

Biologic Plausibility

Although it is generally accepted that gastric acid facilitatescalcium solubility and the related absorption, studies on theeffects of hypochlorhydria on calcium absorption have been

D. A. Johnson (*)Division of Gastroenterology, Eastern Virginia Medical School,885 Kempsville Road, Suite 114,Norfolk, VA 23505, USAe-mail: dajevms@aol.com

Curr Gastroenterol Rep (2010) 12:167–174DOI 10.1007/s11894-010-0103-6

equivocal [4, 5]. Furthermore specific studies of the effectson PPI therapy and calcium absorption have not shownconsistent effect on decreasing the calcium absorption [6, 7].Reduction in urinary calcium excretion was evident in onestudy of healthy volunteers taking omeprazole for 3 dayswhile receiving a meal containing supplemental calcium.Paradoxic to the contention that PPIs might impair bonecomposition is that vaculoar proton pumps, similar instructure to those found in the stomach, are found inosteoclasts and thereby facilitate bone resorption/turnover.Some experimental studies have indicated that PPI therapycan inhibit bone resorption (potentially through inhibitionof proton pumps in osteoclasts), which would thereby beprotective against fractures [8, 9].

Assessment of PPI-Fracture Studies: Evidence Base

To date, all of the reported studies have been case-controlled studies. A study from Denmark evaluated allsubjects sustaining a fracture during the year 2000 [1].Multivariable logistic regression analysis was used to assessthe association between any fracture and exposure variables,with ORs adjusted for alcoholism, employment, comorbid-ities, medications, hospital days, doctor visits, living alone,prior fracture, education level, and income. Among subjectsusing PPIs within the past year, a dose-response relationshipwas not seen for total dose or the mean daily dose (< 0.25daily doses/day OR=1.20, 1.14–1.27 vs 1 daily dose/dayOR=1.11, 1.05–1.18). Adjusted ORs specifically for hipfracture and PPI use were 1.45 (1.28–1.65) for PPI within thepast year and 1.08 (0.94–1.23) for PPI > 1 year earlier. PPIuse within the past year was also associated with spinefractures (adjusted OR=1.60, 1.25–2.04) but not forearmfractures (adjusted OR=0.95, 0.82–1.11).

A nested case-control study from the United Kingdomevaluated patients from the UK General Practice Re-search Database who were at least 50 years of age [2].They also used multivariable logistic regression to assessassociation of hip fractures with the primary exposure ofPPI therapy > 1 year, adjusted for age, sex, body massindex, medications, comorbidities, and prior fracture. Aseparate analysis was done for H2RA users. The adjustedOR was 1.44 (1.30–1.59) for > 1 year of PPIs and 1.23(1.14–1.39) for > 1 year of H2RAs. The duration of PPItherapy and the average daily dose were both significantlyassociated with fracture risk with the adjusted OR for > 1.75times the average daily PPI dose for > 1 year rising to 2.65(1.80–3.90). Of particular note, there was considerablestratification bias in the PPI group for medication use, whichmay predispose to bone density loss: in particular, increaseduse of antihypertensive medications (thiazides), antiseizuremedications, benzodiazepines, and a lower use of estrogens.

A third retrospective case-control study using the Man-itoba Health administrative database assessed cases ofvertebral, wrist, or hip fractures in subjects ≥50 years ofage between 1996 and 2004 [3]. The authors excludedpatients who used osteoprotective medications in the prioryear and residents of long-term care facilities (theirmedications are not accurately tracked in the database).Conditional logistic regression models were used to adjustfor confounding variables (area of residence, income,comorbidity, home health services, medications that affectbone metabolism, or risk of falls) and assessed exposureintervals from 1 to 7 years in yearly intervals. A significantassociation was not shown for fractures and continuous PPIuse (> 70% of time) except in those with 7 years ofcontinuous PPI use (OR=1.92, 1.16–3.18).

The most recent report—a cross-sectional and longitudi-nal case-control study—used the Manitoba Bone MineralDensity Database to determine the relationship with PPIand osteoporosis [10]. In the cross-sectional study, about5,800 patients with osteoporosis were matched to threecontrols with normal bone density. In the longitudinalanalysis, the change in bone density among PPI users andnon-users was assessed sequentially for 2,549 patients whohad sequential standard dual energy x-ray absorptiometry(DEXA scanning). The results demonstrated that PPI usewas not associated either with having osteoporosis of the hip(OR=0.84, 0.55–1.34) or lumbar spine (OR=0.79, 0.59–1.06) for patients with extended PPI use defined as > 1,500doses over 5 years. Additionally, in the longitudinal study,there was no significant decrease in bone mineral density asassessed by DEXA scanning.

Assessment of PPI-Fracture Studies: Conclusions

The earlier published reports linking PPI use to develop-ment of hip fractures were observational studies case-control studies, and thereby have greater potential for biasand therefore less accurate estimates. Additionally, thestrength of association in the PPI studies has been of lowmagnitude. The adjusted summary ORs were 1.18 (95% CI,1.12–1.43), 1.44 (1.30–1.59), and 0.99 (0.90–1.11), but1.92 (1.16–3.18) if 7 years of continuous PPI use [1–3].Given that the estimates and even the upper bounds of mostof the 95% CIs of the OR were well below 2, a strongpossibility exists that these differences could have resultedfrom the bias inherent in observational studies [11••]. Themost recent report does not support the prior reportedassociation [10]. Given that the biologic plausibility is alsonot consistent for causality, it seems reasonable to concludethat overall, it is unlikely that PPI use has a significant riskfor bone density loss and related complications of osteopo-rotic related fractures. Accordingly, the data on bone

168 Curr Gastroenterol Rep (2010) 12:167–174

density loss/osteoporotic fractures would not support thatPPI therapy be discontinued in patients taking PPIs forappropriate indications at appropriate doses.

For the future, it is highly unlikely that randomized,controlled, clinical trials can be accurately done to furtheraddress the question of bone density loss and PPI therapy.The relative rarity of the event for fracture across thepopulation and the extent of potential confounding varia-bles would make conducting this study extremely difficult.It is hoped that more carefully designed cross-sectionalstudies, such as done by Targownik et al. [10], will behelpful in finalizing the answer to this controversy.

PPI and Related Impairment of Antiplatelet Therapy(Clopidogrel)

Scientific Hypothesis for Causality

What factors affect the pharmacodynamics of the clopidog-rel response? Inadequate levels of platelet inhibition withclopidogrel treatment result from intersubject variability inpolymorphisms of CYP2C19 [12, 13]. Furthermore,CYP2C19–and CYP3A4-dependent metabolism of clopi-dogrel has been shown to lead to variability in clinicalresponse. Because it is well known that PPIs are metabo-lized by CYP2C19 and CYP3A4, it has been hypothesizedthat competition at CYP2C19 sites may result in reducedbiologic effects of clopidogrel when co-administered withPPIs. The concept of PPIs interfering with clopidogrelbiotransformation stemmed from in vitro studies thatdemonstrated a pharmacodynamic interaction which wasan attenuated antiplatelet effect as measured by adenosinediphosphate (ADP)–induced platelet aggregation and ele-vated platelet activity [14].

Biologic Plausibility

Do PPIs have an effect on the pharmacodynamic efficacy ofclopidogrel? Initial studies on concomitant PPI and clopidog-rel use presented conflicting evidence regarding PPI interfer-ence with the antiplatelet activity of clopidogrel. Arandomized, open-label, crossover study of the effects oflansoprazole, 30 mg, on the pharmacokinetics and pharmaco-dynamics of single doses of prasugrel, 60mg, and clopidogrel,300 mg, in 24 healthy volunteers found a significant reductionin inhibition of platelet aggregation 24 hours after clopidogreldosing as measured by one marker—5 µM ADP-inducedplatelet aggregation (P<0.05)—but not at other time points,for other markers of platelet inhibition (ie, 20 µM ADP-induced platelet aggregation), or with prasugrel [15]. Gilardet al. [16] initially reported higher platelet reactivity index, asurrogate marker of cardiovascular events, in patients taking

clopidogrel and a PPI versus those taking clopidogrel only.Subsequently, in a double-blind, placebo-controlled, random-ized trial of 124 patients undergoing coronary artery stentimplantation, Gilard et al. [17] reported a significantly higherthrombosis rate (P<0.0001) in patients on omeprazole20 mg/d versus placebo, based on the platelet reactivityindex on day 7 of treatment with clopidogrel (300 mgloading dose followed by 75 mg/d). In addition, a signifi-cantly higher proportion (P<0.0001) of patients in theomeprazole group (60.9%) were poor responders to clopi-dogrel compared with patients in the placebo group (26.7%).The results of these two studies suggested evidence thatomeprazole and lansoprazole decrease the platelet inhib-itory effect of clopidogrel, but the clinical impact of theresults remained uncertain. In a prospective crossoverstudy in 21 patients with coronary heart disease (level-3evidence), Moceri et al. [18] reported a 32%±24% loss inplatelet reactivity (P<0.001) with concomitant esomepra-zole, 20 mg/d, and clopidogrel therapy, 75 mg/d, plus low-dose aspirin, 160 mg/d. The prevalence of low respondersto clopidogrel (defined as platelet reactivity <20%) after7 days of treatment was eightfold higher during concom-itant clopidogrel and esomeprazole treatment. Moreover,the antiplatelet response was improved when the clopi-dogrel dose was increased to 150 mg/d or esomeprazoletherapy was discontinued.

Although some studies have reported decreased anti-platelet activity with concomitant PPI and clopidogreltherapy, others have reported an effect for one PPI andnot another or the absence of an effect. In a nonrandomizedtrial of 300 patients with coronary artery disease, Siller-Matula et al. [19] reported no difference in the plateletreactivity index in patients on clopidogrel, 75 mg/d, andaspirin, 100 mg/d, taking pantoprazole or esomeprazolecompared with those who were not taking a PPI. Aprospective, observational study of 1,000 patients withprevious stent placement and chronic dual antiplatelettreatment with aspirin and clopidogrel, 75 mg/d, who wereundergoing coronary angiography found that reduction inplatelet inhibition was evident with concomitant use ofomeprazole but not with esomeprazole or pantoprazole[20]. Finally, a post hoc analysis of the PRINCIPLE-TIMI 44trial (Prasugrel in Comparison to Clopidogrel for Inhibition ofPlatelet Activation and Aggregation-Thrombolysis In Myo-cardial Infarction), a double-blind, two-phase, crossover studycomparing high-dose clopidogrel (600-mg loading dosefollowed by a daily dose of 150 mg) with prasugrel, a newthienopyridine with more potent antiplatelet effect (60-mgloading dose followed by daily dose of 10 mg), and theTRITON-TIMI 38 trial, a double-blind, randomized studycomparing clopidogrel (300-mg loading dose followed by adaily dose of 75 mg) with prasugrel (60-mg loading dosefollowed by daily dose of 10 mg), found that PPI treatment

Curr Gastroenterol Rep (2010) 12:167–174 169

(taken as a concomitant medication at the discretion of thetreating physician and not as part of the study protocol)reduced the inhibition of platelet aggregation of clopidogreland, to a lesser extent, the antiplatelet effect of prasugrel, butdid not affect overall clinical outcomes of cardiovasculardeath, myocardial infarction (MI), or stroke [21••].

Taken together, the results from these studies imply apharmacodynamic interference of PPI treatment withclopidogrel. The magnitude of the inhibitory effect appearsto be dependent on the PPI used [22•]. The authorsinterpreted the findings to suggest that patients need notavoid concomitant use of PPIs when clinically necessarywhen taking clopidogrel [21••].

Evidence Base

Does concomitant clopidogrel and PPI therapy increase theincidence of adverse cardiac events? The results from twostudies, the Clopidogrel for the Reduction of Events DuringObservation (CREDO) trial [23] and the ClopidogrelMedco Outcomes Study (C-MOS) [24], highlight theconflicting data reported on the possible interactionbetween clopidogrel and PPIs. Post hoc subgroup analysesof patients undergoing or at high likelihood of undergoingpercutaneous coronary intervention (PCI) in the CREDOtrial (176 randomly assigned to receive clopidogrel whowere taking a PPI, 877 randomly assigned to receiveclopidogrel who were not taking a PPI, 190 randomlyassigned to receive placebo who were taking a PPI, and 873randomly assigned to receive placebo who were not takinga PPI) revealed that clopidogrel treatment increased thelikelihood of being free from coronary adverse events(AEs) at 1 year to a similar degree regardless of whether ornot patients were taking a PPI (OR=1.633; 95% CI, 1.015–2.627; P=0.043) [25]. Although the CREDO study was arandomized, placebo-controlled trial, the post hoc nature ofthe present analysis and that patients were not randomizedto receive or not receive PPI treatment suggests concom-itant use of clopidogrel and a PPI is not associated with anincrease in AEs.

By contrast, a retrospective study of 14,383 patients inthe C-MOS database [26] reported that the risk for a majoradverse cardiovascular event (MACE) after coronarystenting was significantly higher in patients who had takena PPI (omeprazole [25.1%; HR=1.39; 95% CI, 1.22–1.57;P<0.0001], esomeprazole [24.9%; HR=1.57; 85% CI,1.40–1.76; P<0.0001], pantoprazole [29.2%; HR=1.61;95% CI, 1.41–1.88; P<0.0001], or lansoprazole [24.3%;HR=1.39; 95% CI, 1.16–1.67; P=0.0004]) compared withthose who had not taken a PPI (17.9%). The authorsconcluded that, although the study provides evidence foran association between PPI use and an increased risk of

MACE, it is not known whether newer PPIs (eg, rabeprazole,dexlansoprazole) would exhibit a similar association [24].C-MOS is the largest trial to date to investigate outcomes inpatients taking clopidogrel and a PPI, but is limited by itsretrospective study design.

Similar to the results from the subgroup analysis of theCREDO trial, PRINCIPLE-TIMI 44 trial, and the TRITON-TIMI 38 trial, other studies have demonstrated a similarincidence of AEs between patients taking clopidogrel and aPPI concomitantly and patients taking clopidogrel alone[22–28]. A retrospective analysis of 1,506 consecutivepatients who underwent PCI and stent implantation inves-tigated the effect of combination PPI use and dualantiplatelet therapy on thromboembolic events [29]. Thisanalysis showed that all-cause mortality and stent restenosisover a median follow-up of 723 days did not differsignificantly between patients on antiplatelet therapy aloneand those on concomitant antiplatelet and PPI therapies.The authors concluded that combination antiplatelet andPPI therapy did not have a significant effect on thrombo-embolic events. Further support for the safe use ofcombination antiplatelet and PPI therapy also comes from aprospective, consecutive cohort study of 673 patients under-going PCI with drug-eluting stent implantation treated withstandard dual antiplatelet therapy of aspirin/clopidogrel for12months and concomitant PPI therapy (lansoprazole 15mg/d,pantoprazole 20 mg/d, or omeprazole 20 mg/d) for 12 months[29]. The 1-year incidence of all-cause mortality, MACE, anddefinite stent thrombosis were 2.2%, 6.2%, and 1.3%,respectively. The authors concluded that the association ofPPI and dual antiplatelet therapy was safe because stentthrombosis occurred at a similar rate to that reported in previousstudies of patients on dual antiplatelet therapy without PPItherapy., In a 1-year follow-up study of in-hospital death in3,059 patients enrolled in the French Acute non-ST–or ST-elevation Myocardial Infarction (FAST-MI) registry andhospital survivors, 1,922 patients were prescribed a PPI, ofwhom 88% also received concomitant clopidogrel [29]. Thisstudy revealed that early PPI use was not correlated within-hospital death in patients taking or not taking clopidogrel(OR=0.91; 95% CI, 0.56–1.48). After adjustment, no correla-tion was found between receiving a prescription for a PPIat discharge and death or MACE over 1 year. In addition,differences in the risk of in-hospital or 1-year mortality werenot observed based upon PPI prescribed (omeprazole versusother PPIs). Additional analyses from the FAST-MI studypopulation that assessed genetic determinants of response toclopidogrel therapy found no effect of PPI therapy on theclinical response to clopidogrel (risk of all-cause mortality,nonfatal stroke, or MI over 1 year) [30].

At the Transcatheter Cardiovascular Therapeutics 2009meeting in San Francisco, CA, Bhatt et al. [31••] presented

170 Curr Gastroenterol Rep (2010) 12:167–174

the results of the COGENT trial (Clopidogrel and theOptimization of Gastrointestinal Events), a randomized,double-blind, double-dummy, placebo-controlled, parallel-group, phase 3 efficacy and safety study of CGT-2168, afixed-dose combination of clopidogrel, 75 mg, and ome-prazole, 20 mg, compared with clopidogrel, 75 mg, alone.Gastrointestinal (GI) and cardiac end points and adjudicatedAEs were evaluated for the 3,627 patients enrolled (abovethe initial target of 3,200). The trial was terminated earlybecause of sponsor bankruptcy; median follow-up was133 days (maximum 362 days) The baseline characteristicsfor cardiovascular risks, nonsteroidal anti-inflammatorydrug (NSAID) use, and Helicobacter pylori status werewell balanced between the omeprazole treatment andplacebo groups. No difference was seen between treatmentgroups in the incidence of MI (HR=0.96; 95% CI, 0.59–1.56) or revascularization (HR=0.95; 95% CI, 0.59–1.55);however, the incidence of GI events was significantly lowerin the combination (omeprazole/clopidogrel) arm (HR=0.55; 95% CI, 0.36–0.85; P=0.007). The authors concludedthat these data provide evidence that no clinically relevantadverse cardiovascular interaction occurred between clopi-dogrel and PPIs over the time period studied.

Most recently, at the 2009 Annual Scientific Sessions ofthe American Heart Association (AHA), Sweeny et al. [32]reported the results of their retrospective cohort study of8,311 consecutive patients who underwent PCI with drug-eluting stent placement. Among patients taking PPIs(n=1,385), the mortality rate after a mean follow-up of2 years was 53.5 deaths/1,000 person-years, significantlyhigher than that of patients not taking PPIs (33.0 deaths/1,000 person years; P<0.001). After multivariate adjust-ment, increased mortality rates were observed for patientsreceiving omeprazole (HR=1.72; 95% CI, 1.11–2.68) orpantoprazole (HR=1.54; 95% CI, 1.13–2.10), but not forthose receiving esomeprazole (HR=0.97; 95% CI, 0.60–1.57)or lansoprazole (HR=1.02; 95% CI, 0.71–1.47) versus thosenot taking a PPI.

To further examine potential adverse cardiovascular andGI outcomes associated with concurrent use of PPIs andclopidogrel, a retrospective cohort study of 20,596 patients(age≥30) who received clopidogrel after being hospitalizedfor myocardial infarction, coronary revascularization, orunstable angina; of these patients, 7,593 (37%) receivedconcurrent PPI therapy [33]. The primary endpoints werehospitalizations for GI bleeding or serious cardiovasculardisease complications (myocardial infarction or suddencardiac death, stroke, or other cardiovascular-related death)during the 7-year study period.

The adjusted incidence of hospitalization for GI bleedingwas lower for patients who received clopidogrel with PPItherapy than for those who received clopidogrel without

PPI therapy (HR, 0.50; 95% CI, 0.39–0.65). Also,concurrent PPI therapy was not associated with significantexcess risk for adverse cardiovascular effects. Of note, forpatients considered to be at highest risk for GI bleeding,concurrent PPI therapy was associated with an absolutereduction of 28.5 (95% CI, 11.7–36.9) hospitalizations forbleeding per 1,000 patient-years

Although limited by its retrospective design, this studyprovides further evidence that the combined use of PPIsand clopidogrel is safe and effective for patients with heartdisease who are deemed to be at high risk for GIcomplications and is not associated with potentially moreformidable cardiovascular complications. Even though allPPIs were analyzed for potential risk-associated events, themajority of patients (62%) received pantoprazole. Given thesmall event rates for the other PPIs, weighted recommen-dations for specific PPIs would not be appropriate.

The results of the studies summarized here provideconflicting evidence both for and against an increasedincidence of cardiovascular events when clopidogrel andPPI therapies are coadministered. Recognizably, all thestudies suggesting potential harm are retrospective claims-based analyses, which are unable to adjust accurately forstratified risk variances. Furthermore, the ORs for harm arelow (most<2), suggesting that there is a strong possibilitythese differences could have been due to the bias inherent inobservational studies. However, there are now four prospec-tive, randomized, controlled trials that have failed to show aclinically relevant effect of concomitant clopidogrel and PPItherapy on cardiovascular outcomes [20, 21••, 24, 31••], andone study by Bhatt et al. [31••] that also evaluated GIconsequences showed a 47% risk reduction for patientstaking omeprazole with clopidogrel.

Further support for a lack of clinically significantadverse interaction comes from the data on genetic poly-morphisms of clopidogrel metabolism. It is apparent fromthe literature that the antiplatelet effect of clopidogrel andinfluence on cardiac adverse events was clearly associatedwith genetic polymorphisms that affect the bioactivation forthe conversion of the prodrug [30, 34]. Patient variation inthe antiplatelet activity is at least in part clearly associatedwith their genetic determinant of the CYP2C19 reducedfunction allele. Recognizably, clopidogrel requires transfor-mation into the active metabolite by the CYP enzymes forits antiplatelet effect. The genes encoding the CYP enzymesare polymorphic, with common alleles conferring reducedfunction. A normal or even rapid biotransfomer wouldachieve the intended prodrug activation, whereas the slowbiotransformer phenotype would not have the samecapabilities and hence some of the drug would bemetabolized and excreted without the bioactivation fromthe prodrug. It was apparent that up to one third of patients

Curr Gastroenterol Rep (2010) 12:167–174 171

were slow biotransformers and this genetic phenotype wascorrelated with adverse cardiac events [30, 34]. Interestingly,however, in one of these two studies, the authors adjusted forexposure to PPIs and found no significant additional adverseeffect, irrespective of the genetic CYP2C19 phenotype [30].Hence, even for the slow converter phenotype with a cleargenetic risk for adverse cardiac events due to lessenedclopidogrel biotransformation, there was no incrementaldifference when corrected for PPI exposure.

National Society Guideline Recommendations RegardingCombination PPI/Clopidogrel Therapy

At present, no definitive evidence establishes that concom-itant use of PPIs and clopidogrel diminishes clopidogrel’sability to prevent cardiac events. After the 2008 AnnualScientific Sessions of the AHA in which the results ofsubgroup analyses from the CREDO trial and initial resultsfrom C-MOS were presented, the American College ofCardiology, American College of Gastroenterology, andAHA issued a joint statement advising that patients shouldcontinue on their current medication regimen unless theirhealth care provider advises otherwise [35]. Since then,more detailed results from C-MOS have been reported,prompting the Society for Cardiovascular Angiography andInterventions to recommend that health care providersconsider prescribing a histaminergic (H2) blocker orantacid instead of a PPI for poststenting patients who areon dual-antiplatelet therapy [36]. Similarly, the US Foodand Drug Administration (FDA) cautioned health careproviders to consider reevaluating the need to continue orstart treatment with a PPI in patients taking clopidogrel[37], and the European Medicines Agency issued a publicstatement recommending that the information for allclopidogrel-containing products be amended to discourageconcomitant use of PPI and clopidogrel-containing medi-cines unless necessary [38]. In addition, after the 2009Annual Scientific Sessions of the AHA, the FDA issuedan advisory that warned health care professionals aboutthe decreased anticlotting efficacy of clopidogrel whengiven with omeprazole, citing new information in healthyvolunteers from data that are not yet published or peerreviewed [38]. Furthermore, the FDA suggested that thispotential interaction was not avoided by separating thedoses of the PPI and clopidogrel (although the dose usedin the cited unpublished information was omeprazole,80 mg—four times the normal dose) [39]. Althoughsocieties and agencies appear to be taking a cautionaryapproach to concomitant PPI/clopidogrel therapy, they arenot necessarily considering whether PPI alternatives, suchas H2 blockers and antacids, will provide similar levels ofeffectiveness for gastroprotection with concomitant anti-platelet therapy as that provided by PPIs.

PPIs and Aspirin: Potential Interaction

A recent report suggests that PPIs may also be associated witha reduced platelet response to aspirin, as evidenced byincreased residual platelet aggregation and platelet activation[40] Investigators from Denmark evaluated 54 patients onPPIs and aspirin, 75 mg/d, from a total group of 418 patientson aspirin with stable coronary artery disease. Patients onconcomitant PPIs had higher platelet aggregation (P=0.003),soluble serum P-selectin(a cell adhesion molecule stored inthe alpha granules of platelets) levels (P=0.005), and higherserum thromboxane B2 levels (P=0.01), all of whichsuggested that these patients may have a reduced antiplateletcardiovascular protective effect. The proposed method ofaction for this altered effect was that an increase inintragastric pH caused by PPIs results in reduced lip-ophilicity of aspirin. Under physiologic acidic conditions,aspirin is absorbed in its lipid state by passive diffusionacross the gastric mucosal membrane according to the pHpartition hypothesis [41]. The authors suggest that if the pHpotentially rises above the pKa (3.5) of acetylsalicylic acid,there may be a pronounced reduction in the lipophilicity ofaspirin [42] . According to previous studies, such chemicalchanges might compromise the bioavailability and therapeu-tic efficacy of aspirin [43, 44]. This study did not present anydata on adverse cardiovascular clinical outcome differencesassociated with their observations of platelet function differ-ences. Accordingly, further study is warranted to investigatepotential causality for adverse effects.

Conclusions

So what data did we have to suggest that this pharmacody-namic observation might not be clinically relevant? A similarinference was made a few years ago about the potentialadverse combination of statin drugs and clopidogrel for thesame reasons of competitive interactions on the P450pathway. When further analyzed in more than 15,000 patientsin the CHARISMA trial (Clopidogrel for High Atherothrom-botic Risk and Ischemic Stabilization, Management, andAvoidance), no clinically apparent adverse cardiac eventswere evident relative to this exposure [45].

The current prospective controlled study data providesstrong reassurance that no clinically relevant adversecardiovascular interaction exists between clopidogrel andPPIs. Additionally, these results call into question the exactrelationship between ex vivo platelet assays and clinicaloutcomes, especially with respect to assessing drug inter-actions. Clearly, platelet assays and observational data maybe factual but are not always accurate for extrapolation toclinical care—in particular, as evident with what has occurredrecently regarding the prescribing recommendations for PPI

172 Curr Gastroenterol Rep (2010) 12:167–174

and clopidogrel use. Platelet assays and observational data arenot substitutes for randomized controlled data.

As with any medication, the key is appropriate use.Health care providers, however, should be attuned toprovide PPI therapy for appropriate symptoms or signs ofacid-related disease and for asymptomatic patients whomeet appropriate risk stratification to justify GI prophylaxis(co-therapy with a PPI) for NSAID/antiplatelet use. It ishoped that these new studies will allow a return to thedirection in which new consensus guidelines by theAmerican College of Cardiology, American College ofGastroenterology, and AHA were moving: toward riskreduction for GI injury in patients at risk [46••].

Prospective clinical trial data will be needed to refute thecurrent trials, which suggest there are no associated significantcardiovascular adverse events due to PPI and clopidogrel.Clearly, federal agency guidelines against the combined use ofPPIs and clopidogrel need to be reanalyzed based on thecurrent clinical evidence.

Disclosure Dr. Johnson has received research support from Astra-Zeneca and Takeda; has served as a consultant to AstraZeneca,Novartis, Eisai, and XenoPort; and has served on the adjudicationcommittee for Eisai. No other potential conflict of interest relevant tothis article was reported.

References

Papers of particular interest, published recently, have beenhighlighted as:• Of importance,•• Of major importance

1. Vestergaard P, Rejnmark L, Mosekilde L: Proton pump inhibitors,histamine H2 receptor antagonists, and other antacid medicationsand the risk of fracture. Calcif Tissue Int 2006, 79:76–83.

2. Yang YX, Lewis JD, Epstein S, et al.: Long-term proton pumpinhibitor therapy and risk of hip fracture. JAMA 2006, 296:2947–2953.

3. Targownik LE, Lix LM, Metge CJ, et al.: Use of proton pumpinhibitors and risk of osteoporosis-related fractures. CMAJ 2008,179:319–326.

4. Bo-Linn GW, Davis GR, Buddrus DJ, et al.: An evaluation of theimportance of gastric acid secretion in the absorption of dietarycalcium. J Clin Invest 1984, 73:640–647.

5. Recker RR: Calcium absorption and achlorhydria. N Engl J Med1985, 313:70–73.

6. O'Connell MB, Madden DM, Murray AM, et al.: Effects of protonpump inhibitors on calcium carbonate absorption in women: arandomized crossover trial. Am J Med 2005, 118:778–781.

7. Serfaty-Lacrosniere C, Wood RJ, Voytko D, et al.: Hypochlorhy-dria from short-term omeprazole treatment does not inhibitintestinal absorption of calcium, phosphorus, magnesium or zincfrom food in humans. J Am Col Nutr 1995, 14:364–368.

8. Graziani G, Como G, Badalamenti S, et al.: Effect of gastric acidsecretion on intestinal phosphate and calcium absorption in

normal subjects. Nephrol Dialysis Transplant 1995, 10:1376–1380.

9. Tuukkanen J, Vaananen HK: Omeprazole, a specific inhibitor of H+-K+-ATPase, inhibits bone resorption in vitro. Calc Tissue Int 1986,38:123–125.

10. Targownik LE, Lix LM, Leung S, Leslie WD: Proton pumpinhibitor use is not associated with osteoporosis or acceleratedbone mineral density loss. Gastroenterology 2010, 138:896–904.

11. •• Laine L: Proton pump inhibitors and bone fractures? Am JGastroenterol 2009, 104:S21–S26. This article is an excellentoverview of the literature and evidence-based assessment ofrelative risks.

12. Varenhorst C, James S, Erlinge D, et al.: Genetic variation ofCYP2C19 affects both pharmacokinetic and pharmacodynamicresponses to clopidogrel but not prasugrel in aspirin-treatedpatients with coronary artery disease. Euro Heart J. 2009,30:1744–1752.

13. Shuldiner AR, O’Connell JR, Bliden KP, et al.: Association ofcytochrome P450 2C19 genotype with the antiplatelet effect andclinical efficacy of clopidogrel therapy. JAMA. 2009, 302:848–858.

14. Gilard M, Arnaud B, Cornily JC , et al.: Influence of omeprazoleon the antiplatelet action of clopidogrel associated with aspirin:the randomized, double-blind OCLA (omeprazole clopidogrelaspirin) study. J Am Coll Cardiol 2008, 51:256–260.

15. Small DS, Farid NA, Payne CD, et al.: Effects of the proton pumpinhibitor lansoprazole on the pharmacokinetics and pharmacody-namics of prasugrel and clopidogrel. J Clin Pharmacol 2008,48:475–484.

16. Gilard M, Arnaud B, Le Gal G, et al.: Influence of omeprazole onthe antiplatelet action of clopidogrel associated with aspirin[letter]. J Thromb Haemost 2006, 4:2508–2509.

17. Gilard M, Arnaud B, Cornily J-C, et al.: Influence of omeprazoleon the antiplatelet action of clopidogrel associated with aspirin. JAm Coll Cardiol 2008, 51:256–260.

18. Moceri P, Drogoul L, Ferrari E: Strong decrease of the clopidogrelantiplatelet effect with esomeprazole, but not with ranitidine andrecovering of the antiplatelet effect by doubling the dosage ofclopidogrel [abstract 895]. Euro Heart J 2009, 30(Suppl):150.

19. Siller-Matula JM, Spiel AO, Lang IM, et al.: Effects ofpantoprazole and esomeprazole on platelet inhibition by clopi-dogrel. Am Heart J 2009, 157:148.e1–148.e5.

20. Sibbing D, Morath T, Stegherr J, et al.: Impact of proton pumpinhibitors on the antiplatelet effects of clopidogrel. ThrombHaemost 2009, 101:714–719.

21. •• O’Donoghue ML, Braunwald E, Antman EM, et al.: Pharma-codynamic effect and clinical efficacy of clopidogrel andprasugrel with or without a proton-pump inhibitor: an analysisof two randomised trials. Lancet 2009, 374:989–997. This is a keyarticle of two prospective trials with combined clopidogrel andPPI usage and assessment of cardiac outcomes. Although the PPIuse was not an a priori risk assessment (and only one third of thestudy cohort took these), there is prospective data with excellentbalance of stratified risks between groups.

22. • Sibbing D, Kastrati A: Risk of combining PPIs with thienopyr-idines: fact or fiction? Lancet 2009, 374, 952–954. This article isa good summary editorial of the two prospective trials withclopidogrel and patients who also were taking PPIs.

23. Dunn SP, Macaulay TE, Brennan DM, et al.: Baseline protonpump inhibitor use is associated with increased cardiovascularevents with and without the use of clopidogrel in the CREDO trial[abstract 3999]. Circulation 2008, 118(Suppl 2):S815.

24. Stanek EJ, Aubert RE, Flockhart DA, et al.: A national study ofthe effect of individual proton pump inhibitors on cardiovascularoutcomes in patients treated with clopidogrel following coronarystenting: the Clopidogrel Medco Outcomes Study [late-breakingclinical trial O-11]. Presented at the 2009 Annual Scientific

Curr Gastroenterol Rep (2010) 12:167–174 173

Session of the Society for Cardiovascular Angiography andInterventions. Las Vegas, NV; May 6–9, 2009.

25. Dunn SP, Macaulay TE, Brennan DM, et al.: Baseline protonpump inhibitor use is associated with increased cardiovascularevents with and without the use of clopidogrel in the CREDO trial[abstract 3999]. Circulation 2008, 118(Suppl 2):S815.

26. Aubert RE, Epstein RS, Teagarden JR, et al.: Proton pumpinhibitors effect on clopidogrel effectiveness: the ClopidogrelMedco Outcomes Study [abstract 3998]. Circulation 2008, 118(Suppl 2):S815.

27. Jarai R, Brosovic I, Tentzeris I, et al.: Effects of proton-pumpinhibitors on outcome of patients discharged on dual-antiplatelettherapy after percutaneous coronary intervention and stentimplantation [abstract P675]. Euro Heart J 2009, 30(Suppl):98.

28. Rossini R, Nijaradze T, Musumeci G, et al.: Safety of proton-pump inhibitors in patients on oral antiplatelet therapy withaspirin and clopidogrel after DES implantation [abstract P4532].Euro Heart J 2009, 30(Suppl):741.

29. Simon T, Quandalle PH, Machecourt J, et al.: Proton pumpinhibitors and clopidogrel response on cardiovascular majorevents inpatients after acute myocardial infarction. Data from theFAST-MI registry of the French Society of Cardiology [abstractP2121]. Euro Heart J 2009, 30(Suppl):335.

30. Simon T, Verstuyft C, Mary-Krause M, et al.: Genetic determi-nants of response to clopidogrel and cardiovascular events. NEngl J Med 2009, 360:363–375.

31. •• Bhatt DL, Cryer B, Contant CF, et al., on Behalf of theCOGENT Investigators: The COGENT trial. Presented at theTranscatheter Cardiovascular Therapeutics Meeting. San Fran-cisco, CA; September 21–25, 2009. This was a late-breakingabstract, so no published citation but a key article is forthcoming,as this will be the first prospective, randomized, controlledclinical trial assessing both cardiac and GI adverse events.

32. Sweeny JM, Kini A, Muntner P, et al.: Mortality associated withproton pump inhibitors following percutaneous coronary inter-ventions with drug eluting stents [abstract 4323]. Circulation2009, 120:S936.

33. Ray WA, Murray KT, Griffin MR, et al.: Outcomes withconcurrent use of clopidogrel with proton pump inhibitors. AnnIntern Med 2010, 152:337–355.

34. Mega JL, Close Sl, Wiviott SD , et al.: Cytochrome P-450polymorphisms and response to clopidogrel. N Engl J Med 2009,360:354–362.

35. American College of Cardiology (ACC)/American College ofGastroenterology (ACG)/American Heart Association (AHA)joint comment on studies regarding possible interaction ofclopidogrel and proton pump inhibitors [news release]. AmericanCollege of Cardiology News Release; November 12, 2008.Available at http://www.acc.org/media/releases/highlights/2008/nov08/ppi.htm. Accessed February 2010.

36. Society for Cardiovascular Angiography and Interventions: SCAIStatement on “A national study of the effect of individual protonpump inhibitors on cardiovascular outcomes in patients treatedwith clopidogrel following coronary stenting: the clopidogrelMedco outcomes study.” Washington, DC: Society for Cardio-vascular Angiography and Interventions; 2009. Available at http://www.scai.org. Accessed February 2010.

37. US Food and Drug Administration: Early communication aboutan ongoing safety review of clopidogrel bisulfate (marketed asPlavix). Available at http://www.fda.gov/Drugs/DrugSafety/PostmarketDrugSafetyInformationforPatientsandProviders/DrugSafetyInformationforHeathcareProfessionals/ucm079520.htm. AccessedFebruary 2010.

38. National Prescribing Centre: EMEA statement on possible interac-tion between clopidogrel and proton pump inhibitors. Available athttp://www.npci.org.uk/blog/?p=354. Accessed February 2010.

39. US Food and Drug Administration: Follow-up to the January 26,2009, early communication about an ongoing safety review ofclopidogrel bisulfate (marketed as Plavix) and omeprazole(marketed as Prilosec and Prilosec OTC). Available at http://www.fda.gov/Drugs/DrugSafety/PostmarketDrugSafetyInformationforPatientsandProviders/DrugSafetyInformationforHeathcareProfessionals/ucm190784.htm. Accessed February 2010.

40. Würtz M, Grove EL, Kristensen SD, Hvas AM: The antiplateleteffect of aspirin is reduced by proton pump inhibitors in patientswith coronary artery disease. Heart (British Cardiac Society)2010, 96:368–371.

41. Shore PA, Brodie BB, Hogben CA: The gastric secretion of drugs:a pH partition hypothesis. J Pharmacol Exp Ther 1957, 110:361–369.

42. Hollander D Dadufalza VD, Fairchild PA: Intestinal absorption ofaspirin. Influence of pH, taurocholate, ascorbate and ethanol. JLab Clin Med 1981, 98:591–598.

43. Giraud MN, Sanduja SK, Felder TB, et al.: Effect of omeprazoleon the bioavailability of unmodified and phospholipid-complexedaspirin in rats. Aliment Pharmacol Ther 1997, 11:899–906.

44. Fernandez-Fenrandez FJ. Might proton pump inhibitors preventthe antiplatelet effects of low-or very low-dose aspirin? ArchIntern Med 2002, 162:2248.

45. Saw J, Brennan DM, Steinhubl SR, et al.: Lack of Evidence of aClopidogrel–Statin Interaction in the CHARISMA Trial. J AmColl Cardiol 2007, 50:291–295.

46. •• Bhatt D, Scheiman J, Abraham N, et al.: ACCF/ACG/AHAExpert Consensus Document on Reducing the GastrointestinalRisks of Antiplatelet Therapy and NSAID use. Am J Gastro-enterol 2008, 103:2890–2907. This article is an excellentevidence-based consensus between the cardiology and GI socie-ties to highlight the need for appropriate risk assessment for GIadverse outcomes and appropriate use of prophylaxis to protectagainst antiplatelet/NSAID GI injury.

174 Curr Gastroenterol Rep (2010) 12:167–174